Log inSubmit server
Rigor MCP logo

Rigor MCP

rigor-mcp MCP server computes assumption-checked statistical tests, power, corrections, and sequential analyses for AI agents.

InstallRepository
Full README Compare alternatives Embed widget Claim this listing

Install Rigor MCP

Pick your client, then add it in one click or paste the config.

STDIOPython
~/Library/Application Support/Claude/claude_desktop_config.json
claude_desktop_config.json
{
  "mcpServers": {
    "rigor-mcp": {
      "command": "uvx",
      "args": [
        "rigor-mcp"
      ]
    }
  }
}

Settings → Developer → Edit Config, merge this into mcpServers, then fully quit and reopen Claude Desktop.

Automated check passed— started and listed 37 tools correctly (5d ago).

Overview

The rigor-mcp MCP server provides computed, cited statistical inference for AI agents, including hypothesis tests, effect sizes, power analysis, corrections, and sequential testing. It exposes decision and batch helpers alongside tools for t-tests, proportions, categorical data, correlations, regression, and non-parametric methods. Results include statistical details, confidence intervals where supported, citations, and assumption warnings. Reach for it when an agent must calculate results from supplied data rather than recall a similar textbook example.

Use cases

  • Run assumption-checked hypothesis tests on supplied sample data
  • Calculate power or required sample size for planned experiments
  • Correct a batch of pairwise comparisons for multiple testing
  • Select a statistical test from data and study characteristics
  • Evaluate repeated dashboard checks with sequential testing

Key features

  • Parametric and non-parametric hypothesis tests
  • Effect-size and confidence-interval calculations
  • Power and sample-size analysis
  • Bonferroni and Benjamini-Hochberg correction
  • Always-valid sequential testing
  • Test selection and batch comparison helpers

Tools37

What Rigor MCP lets your agent do. Expand a tool for its parameters.

Verified live~9.3k tokens of context

Security & health

What it can reach and how reliably it runs — from our own checks, not the maintainer’s claims.

ActiveLast checked 2d ago

Recent health check succeeded.

What it can do

From the 37 tools this server returned when we ran it; 37 declare their own safety hints.

Read-only tools · 37
Network at startup2 connections
Files written at startup30

Observed 2026-10-17 in an isolated sandbox. A description of reach, not a security audit.

Supply-chain signal

No high-severity advisories surfaced by our automated scan.

Critical 0High 0Medium 0Low 0

Scanned 12d ago via OSV.dev · rigor-mcp (PyPI)

Adoption & maintenance

Factual signals from GitHub, npm, and our automated checks — not a rating.

GitHub stars1Stargazers on the source repository.
Last commit1mo agoMost recent push to the default branch.
Availability100%Our rolling endpoint + install checks that succeeded.
Install checkPassedOur sandbox started it and listed its tools.
Tools exposed37Callable tools this server registers over MCP.
Directory activity1 viewsConfig copies, upvotes, and views on AllMCPs.

How Rigor MCP works

What rigor-mcp MCP server does

The rigor-mcp MCP server gives AI agents a collection of statistical inference tools that calculate results from scratch. It covers one- and two-sample, paired, and proportion tests; chi-squared, Fisher's exact, McNemar's, ANOVA, and variance tests; non-parametric alternatives; correlations; simple linear regression; effect sizes; power and sample-size calculations; and multiple-comparisons correction.

Most inference results are returned as structured test results containing a statistic, degrees of freedom when applicable, a two-tailed p-value, a confidence interval, a citation, and warnings about assumptions. The package also includes a test-selection helper and a batch comparison tool, so an agent can choose an analysis or run many group comparisons without manually coordinating every step.

How it works

The rigor-mcp MCP server is implemented as a Python package with an MCP server entry point. Its statistical calculations use the Python standard library rather than third-party numerical dependencies. Distribution functions for t, chi-squared, and F statistics are implemented directly and checked against stated mathematical identities.

Power calculations support one- and two-sample t-tests and two-proportion z-tests. The implementation can calculate power from a sample size or solve for sample size from a target power using the same underlying function in both directions. Effect-size tools include Cohen's d, Hedges' g, Cohen's h, Cramér's V, eta-squared, omega-squared, and rank-biserial correlation.

Sequential tools address analyses checked repeatedly during data collection. The package provides always-valid mSPRT methods for two-sample means and two proportions, as well as a seeded simulation that estimates false-positive inflation from naive repeated peeking. These methods are intended for repeated checks rather than ordinary fixed-sample testing.

Setup and configuration

Install the PyPI distribution with:

Terminal
pip install rigor-mcp

The installation provides both the rigor command-line interface and the rigor-mcp MCP server command. The importable package and CLI name remain rigor, while the distribution name is rigor-mcp. MCP clients can invoke the server with uvx rigor-mcp.

No environment variables or external data-service credentials are documented as required. The package as a whole depends on the official MCP SDK; the statistical modules themselves use the standard library.

Tools and capabilities

The rigor-mcp MCP server includes capabilities for:

  • Selecting an appropriate test from data type, group count, pairing, distribution concerns, association questions, or repeated checking.
  • Running pairwise comparisons across multiple groups and applying Bonferroni or Benjamini-Hochberg correction to the complete batch.
  • Choosing parametric or rank-based methods, including Mann-Whitney U, Wilcoxon signed-rank, and Kruskal-Wallis tests.
  • Computing Pearson or Spearman correlation and testing a single-predictor ordinary least squares regression.
  • Calculating power, required sample size, effect sizes, and corrections for multiple comparisons.

The decision helper can route repeated-checking cases to a sequential method when one exists, or explicitly report that no such route is available.

Limitations and notes

The rigor-mcp MCP server does not replace assessment of whether a study design or data-generating process meets a method's assumptions. Its outputs include warnings, such as reliance on normality with small samples and low expected cell counts, so those warnings must be considered before interpreting a result.

Sequential methods are documented for two independent groups with continuous outcomes or proportions; other repeated-checking situations may not have a corresponding always-valid tool. Regression is limited to simple, single-predictor ordinary least squares. The package also provides calculations and guidance, not a general-purpose data collection or database integration layer.

Read the full README View source on GitHub

Reviews

No reviews yet — be the first to share how it worked for you.

Frequently Asked Questions about Rigor MCP

rigor-mcp is an MCP server that connects AI agents to Python-based statistical inference tools. Its main tools run hypothesis tests, calculate effect sizes and power, correct multiple comparisons, select suitable tests, batch pairwise comparisons, and perform always-valid sequential analyses for means or proportions.

Install it with `pip install rigor-mcp`. The package supplies the `rigor-mcp` MCP command and the `rigor` CLI command. MCP clients can also invoke it with `uvx rigor-mcp`, with no documented environment variables or extra packages to configure.

Yes, rigor-mcp supports Bonferroni and Benjamini-Hochberg corrections. Its batch comparison helper can run pairwise tests across two or more groups and apply a selected correction to the full set of comparisons.

Yes, for two-sample means and two proportions, rigor-mcp provides always-valid sequential tests based on mSPRT. It also includes a seeded simulation that estimates the false-positive inflation caused by repeatedly applying a fixed-sample test while data arrives.

rigor-mcp includes t-tests, proportion z-tests, chi-squared tests, Fisher's exact test, McNemar's tests, one-way ANOVA, Levene's test, Mann-Whitney U, Wilcoxon signed-rank, Kruskal-Wallis, Pearson and Spearman correlation, and simple single-predictor regression.

Related MCP servers

All in Data Science Tools All alternatives
Add a badge to your READMEShow it’s listed — live SVG, Markdown or HTML.

AllMCPs Directory Badge

Full Badge Customizer

Showcase your server listing on GitHub or your project documentation. Embed this dynamic SVG badge to highlight official listing status and live engagement.

Badge Style:
Live Dynamic SVG PreviewRigor MCP AllMCPs Directory Badge
Markdown (GitHub README)
[![AllMCPs](https://allmcps.com/api/badge/rigor-mcp?style=directory)](https://allmcps.com/mcp/rigor-mcp)
HTML Embed
<a href="https://allmcps.com/mcp/rigor-mcp"><img src="https://allmcps.com/api/badge/rigor-mcp?style=directory" alt="Rigor MCP on AllMCPs" /></a>